Research On Apron Surveillance Video Compression Method Based On Deep Learning

Apron surveillance video is an important basis for airport to ensure the safety of apron area.The storage and management of apron surveillance video is of great significance for the airport.The new version of "Security facilities of civil transportation airport"(MH / T7003-2017)stipulates that the storage time of surveillance video in important areas of the airport shall not be less than 90 days,and the quality of playback image shall not be significantly lower than that of real-time image.Due to many surveillance cameras,long storage time and high image quality requirements in apron area,it brings great challenges to the video storage of airport.In order to solve the storage problem of apron surveillance video,this paper proposes a method of apron surveillance video compression based on deep learning.According to the characteristics of apron surveillance video,the moving objects and static objects in the video are stored separately.At the same time,according to the change of illumination in the video,the illumination information in the video is stored.Therefore,the video is compressed into static layer,object layer and illumination layer.The static layer mainly stores the background image of the video.The background image refers to the image composed of still objects in the video.The object layer stores all moving object images and image coordinate information in the video.The illumination layer mainly stores the brightness change and light information in the video.The static layer and object layer are obtained by using the object detection method.The illumination layer uses illumination model to obtain light information.Because a video composed of several image frames is transformed into a background image,a number of object images and light information storage structure,the storage space of video can be greatly reduced.As the apron is an outdoor environment,rainy and foggy weather sometimes occurs.Considering these special weather,the weather layer is added on the basis of static layer,object layer and illumination layer.The weather layer mainly stores the information of rain and fog.About the acquisition and storage of rain and fog,the information of rain and fog in video can be obtained through the algorithm of removing rain and fog.In the video decompression stage,the weather information is restored to the image,so as to restore the real apron monitoring video.In order to verify the performance of the algorithm proposed in this paper,we test the apron surveillance video of different airports,different times and different weather.The experimental results show that this method can achieve good results for the long-time apron surveillance video compression.